AI for healthcare and in silico clinical trials

AI-based methods and software to support In Silico Clinical Trials (ISCT) and decision making in clinical settings.

At RAISE Lab we focus on the definition and analysis of quantitative models for treatment strategies and human physiology & physiologically based pharmacokinetics.

Treatment strategy models

aka virtual doctors

Models of treatment strategies can be regarded as virtual doctors. Indeed they define the decisional process of a clinician while administering a treatment, or the controller policy of a biomedical device.

Human physiology & physiologically based pharmacokinetics models

aka virtual patients

Models of human physiology & physiologically based pharmacokinetics can be regarded as virtual patients. They are quantitative, mathematical models of the human physiology of interest and the kinetics of the relevant pharmaceutical compounds for the use case at hand.

We exploit such models to define patient digital twins, to perform in silico clinical trials (ISCT), what-if analyses, as well as design of optimal personalised treatments and of biomedical devices.

Patient digital twins

Virtual patients whose evolution and reactions to drugs closely matches those of their associated human patient

In silico clinical trials (ISCT)

Clinical experimentations via computer simulations

In silico what-if analyses

Evaluation of different treatment scenarios via computer simulations on patient digital twins

In silico individualised treatment design

Optimisation of personalised treatments via computer simulations on patient digital twins

In silico optimisation of biomedical devices

Performance maximisation of biomedical devices via computer simulations on patient digital twins

More details

ISCT aim at performing, via computer simulations, the typical activities carried out to assess safety and efficacy of pharmacological treatments, biomedical devices, or other therapeutic procedures.

ISCT ask for highly cross- and inter-disciplinary knowledge and methods in: artificial intelligence, formal verification, model checking, computer engineering, modelling, simulation, high performance computing, biology, physiology, pharmacology, omics.

What are the advantages of ISCT?

Being entirely model-based, ISCT have the potential to:

Reduce time and cost of traditional approaches
Prioritise in vivo trials, via selection of most relevant patient phenotypes
Avoid in vivo assessment of unsuccessful designs of drugs/treatments or device design (early pruning)
Reduce number of animals and humans involved
Enable precision medicine, in silico what-if analyses, individualised treatment design and biomedical device optimisations
Tackle areas where human volunteer recruiting is hard or unethical (e.g., rare diseases, paediatric drugs, pregnancy).

From qualitative knowledge to quantitative computational models

ISCT leverage the available qualitative knowledge stemming from biology, omics, patho-physiology and clinical data to define quantitative models of the human physiology of interest and of the PKPD of medicinal compounds (drugs), clinical guidelines, treatment and decision strategies, medical devices.

RAISE

RAISE Research Lab in Artificial Intelligence for Systems Engineering